Back to Community
Issues dealing with NaN

Hello All,

I am trying to deal with leverage, and in setting up the order logic to do so I am running into some issues around NaN values.

What I am trying to do is simply base the order value on the available funds in the portfolio divided by the stock price. The issue was the math with a NaN object. I initially tried to just skip over NaN values:

Here is the code (I have not been able to run a back test because of said errors):

def handle_data(context, data):

    results = pipeline_output('my_pipeline')  
    context.pipeline_results = results.sort_values('SMARatio', ascending=False).iloc[:300]  
    for stocks in context.pipeline_results:  
        cash = context.portfolio.cash  
        stock_value = context.pipeline_results.Close  
        if np.isnan(stock_value) == False:  
            try:  
                #Order Logic  
            except:  
                return  

The Error (if np.isnan(stock_value) == False:):
ValueError: The truth value of a Series is ambiguous. Use a.empty, a.bool(), a.item(), a.any() or a.all().

I get the same error this way:

def handle_data(context, data):

    results = pipeline_output('my_pipeline')  
    context.pipeline_results = results.sort_values('SMARatio', ascending=False).iloc[:300]  
    for stocks in context.pipeline_results:  
        cash = context.portfolio.cash  
        stock_value = context.pipeline_results.Close

        stock_value[np.isnan(stock_value)] = 0  
        if stock_value != 0.0:  
            purchase_target = (cash/stock_value)  

I can convert the NaN values to zeros before any math, but then I have an issue with infinities. So I would prefer to handle NaNs within an IF Statement.